Handle ECONNRESET error when keepAlive connection is terminated by …

[kobalski]

…the server. In a long-running lambda container keep-alive connections are reset by our collector servers due to nginx config for keepalive_timeout is 30 seconds. If a request fails due to this error, we do not log the error but we try to send the data again otherwise we miss monitoring data.

[serkan-ozal]

It is better to try only once. Not sure whether it is possible in practice, but in theory, this might result infinite recursion by calling sendReports recursively. To prevent this:

async sendReports(dontRetry: boolean): Promise<void> {
    ...
    await Promise.all(reportPromises).catch(async (err) => {
         if (dontRetry != true && err.code === 'ECONNRESET') {
             ThundraLogger.getInstance().debug('Connection reset by server. Will send monitoring data again.');
             await this.sendReports(true);
             return;
         }
         ThundraLogger.getInstance().error(err);
    });
}

[kobalski]

I included connectionRetryCount to Reporter to make sure we try to reconnect to server only once.

[coveralls]

Pull Request Test Coverage Report for Build 561

• 0 of 0 changed or added relevant lines in 0 files are covered.
• 24 unchanged lines in 1 file lost coverage.
• Overall coverage decreased (-0.1%) to 69.955%

---

Files with Coverage Reduction	New Missed Lines	%
dist/Reporter.js	24	59.09%

Totals
Change from base Build 556:	-0.1%
Covered Lines:	2242
Relevant Lines:	2977

---

💛 - Coveralls